slope between 20øS and 28øS are investigated for the existence and strength of western boundary currents from near the surface down to the North Atlantic Deep Water. The

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 103, NO. C3, PAGES 5429-5437, MARCH 15, 1998

Direct measurements of western boundary currents off Brazil between 20øS and 28øS

Thomas J. Miiller, • Yoshimine Ikeda, 2 Norbert Zangenberg, • and Luiz V. Nonato 2

Abstract. Current measurements from three moored arrays on the Brazilian continental

slope between 20øS and 28øS are investigated for the existence and strength of western boundary currents from near the surface down to the North Atlantic Deep Water. The

strengthen its volume transport to 16.2 x !06 m3/s. Antarctic Intermediate Water is transported in a well-developed boundary current southward at 28øS and northward north of Cabo Frio (24øS). This result supports earlier suggestions derived from the analysis of hydrographic data that Antarctic Intermediate Water enters the Brazil Basin from the east and bifurcates as it meets the continental break off Brazil. North Atlantic Deep Water is transported southward in a weakly developed boundary current that leads to lower estimates of volume transport than expected from earlier hydrographic data analysis.

1. Introduction

The South Atlantic is known to be the highway on which the major interoceanic exchange of water masses and, conse- quently, of heat and salt occurs. As a result of the Meteor expedition 1925-1927 into the South Atlantic, Wast [!935] and Defant [1936a, b] illustrated the South Atlantic's deep and near-surface circulation, which in its general structure still holds for the subtropical western South Atlantic off Brazil.

Much of the associated volume transports is carried in western boundary currents that in early theories are required to bal- ance the interior ocean circulation [Stommel, 1948; Stommel and Aarons, 1960]. Principally, this concept could apply to all layers of deep and intermediate waters. For the Brazil Basin, one therefore would expect not only the near-surface Brazil Current balancing the wind-driven circulation, but also bound~

ary currents in the layers of the Antarctic Intermediate Water (AAIW), the Upper Circumpolar Deep Water (UCDW), the three layers of the North Atlantic Deep Water (NADW), and

the Antarctic Bottom Water (AABW). While deep boundary

currents carrying NADW southward and AABW northward

have been described in the literature, a northward spreading of

broad reviews of the present knowledge of the South Atlantic's

graphic measurements, the reader is referred to Reid [!989], Peterson and Stramma [199!], DeMadron and Weatherly [1994], and Tsuchiya et al. [!994].

Few direct current observations were available from the

boundary current system along the South American coast when the World Ocean Circulation Experiment (WOCE) began with its Deep Basin Experiment (DBE) in 1990. The then available

records from moored current meters show the existence of a

•Institut ffir Meereskunde an der Universit•it Kiel, Kiel, Germany.

2Instituto Oceanografico da Universidade de Sao Paulo, Sao Paulo,

Brazil.

Copyright 1998 by the American Geophysical Union.

Paper number 97JC03529.

0!48.0227/98/97JC.03529509.00

deep boundary current that carries bottom water of Antarctic origin equatorward as it enters the Argentine Basin north of the Falkland Ridge [Whitworth eta!., 1991], flows along the continental break at 38øS [Weatherly, 1993] and at 35øS [Reid et al., 1977], and partly leaves the basin through the Verna Chan- nel into the Brazil Basin [Reid et al., 1977; Hogg et al., 1982].

Within the level of the NADW, the few direct measurements from outside the tropics show poleward flow across the Santos Plateau [Reid et al., 1977]. Farther south, at 38øS [Weatherly, 1993], the flow is equatorward, which is consistent with the hypothesis that a poleward boundary jet of NADW occurs further north at about 32øS [see Weatherly, 1993].

Within the other three layers, the only reported direct pre-

WOCE current measurements made before 1990 stem from a Pegasus system [Spain et al., 1981] that was used in the two western channels of the Vitoria Trindade Ridge (VT, 20øS, Figure 1), and twice off Cabo Frio (CF, 24øS; Figure 1 [Evans and Signorini, 1985; Miranda et al., 1985]). As expected, in both regions their profiles show the Brazil Current flowing poleward while the AAIW flow was equatorward at speeds up to 35 cm/s.

Only some profiles at the Vitoria Trindade Ridge were deep enough to measure the flow of NADW. These showed pole- ward flow where the sill depth is deep enough to let it pass.

The analysis of hydrographic data supported the hypothesis that at least part of AABW and NADW is transported by continuous western boundary currents along the South Amer-

ican shelf break [see Reid, 1989]. For AAIW, this hypothesis

[1977], and Reid [1989] found the AAIW flowing equatorward as a boundary only until the confluence zone of the Brazil and Malvinas Currents at about 40øS, where it joins the anticyclonic subtropical gyre circulation. Following the gyre circulation, it then enters the Brazil Basin from the east at about 25øS, approaches the continental break just south of Cabo Frio, and

then bifurcates into two branches, one poleward and one equa-

torward along the Brazilian coast. This scheme has been con~

firmed with more data by Warner and Weiss [1992] and Suga and Talley [1995]. Note that for the upper layer, Tsuchiya

5429

15 ø 1• ø

I ....--. C•2 II

25 25 •

30 ø

50ø W 4'5ø 4'0ø 35ø

Figure la. Location of current meter mooring arrays off Brazil: Vitoria Trindade Ridge (VT; see Figure lb for details) at 20øS with moorings BS1 and BS2, November 1989 to No- vember 1990; Cabo Frio (CF) at 24øS with moorings CF1/347, CF2/348, CF3/349, September 1992 to August 1993; WOCE array ACM3 with moorings BW/333, BM/334, and BE/335, January 1991 to November 1992. The 200, 1000, and 2000 m depth contours are indicated.

current meter measurements recently made at the slope show a steady northward flow at the level of the AAIW, which is in agreement with Reid [1989]. Further north, one float was launched into the AA!W core off Salvador at 12øS. That float

northward until it began to move offshore at about 9øS [OIli.

traut et al., 1994].

Also within the DBE, a current meter mooring array was set offshore near the equator [Schott et at., 1993]. While for the

AAIW, even the flow direction is unclear due to coarse spatial

array ACM24. Here, the AAIW flows northward as a thin jet, while the NADW is deflected to the deep basin by the ridge on

its way south.

To further investigate the vertical structure of the western boundary current system down to the NADW off Brazil be- tween 20øS and 28øS, we use the current measurements from three moored arrays (Figure 1, Table !): WOCE array ACM3 at 28øS [Tarbell et at., 1994], data from the array CF across the slope off Cabo Frio at 24øS, and data from two moorings in the

two western channels of the Vitoria Trindade Ridge (VT) at

Zenk [1993], Zenk and Hogg [1996], and Hogg and Zenk [1997].

which the AA!W enters the basin. Whether these cells were

connected at the western boundary by a boundary current remained unclear from his analysis.

Since 1989, several experiments were carried out within or related to the DBE to directly measure currents within the western boundary currents. From a RAFOS float trajectory analysis, it seems unlikely that AAIW flows as a continuous western boundary current along the western slopes between 25øS and 35øS toward the Equator [Boebet et aI., 1997]. How- ever, north of 25øS, Campos et al. [1996] quote that 1 year long

30'

2O ø S

30'

> i i ,

30' 38 ø W 30'

Figure lb. Location of VT array with moorings BS1/330 and BS2/33! in the two western channels of the Vitoria Trindade Ridge. The 200, 1000, and 2000 m depth contours are indi-

cated.

2. Current Meter Data and Methods

Two moorings were set in the two western channels of the Vitoria Trindade Ridge at 20øS during Meteor cruise 11/2 in November 1989 and recovered 1 year later during Meteor cruise 14/3 [tIinz et al., 1991], providing record lengths of up to 368 days. Each had two Aanderaa type current meters in the South Atlantic Central Water (SACW) range, one at the core depth of the AAIW and one about 50 m above the bottom.

Unfortunately, the two upper instruments of the western mooring (BS1/330) were lost (probably due to fishing), and the meters in the AAIW level of both moorings provided data for only 157 days (BS1/330) and 112 days (BS2/331) due to instru-

ment failure.

The CF array was set from N/Oc. Prof. W. Besnard in Sep- tember 1992 and recovered by the same vessel 1 year later in 1993 with record lengths up to 327 days. It consisted of three moorings that were set on a line normal to the slope at 2260 m (CF1/347), 980 m (CF2/348), and a single instrument mooring on the shelf at 210 m water depth (CF3/349) close to its break.

Because of fishing activity, the two uppermost instruments were cut from mooring CF2/348. While the meter at 500 m depth was lost, a fisherman returned the meter from 240 m depth with a 284 day record. Also, the instrument at 920 m of that mooring recorded no data between days 35 and 154 be- cause of a bad piece of tape. To complete the record, the missing 119 days were linearly interpolated between the aver- ages of the two good parts; this procedure does not change the average much but underestimates the variances of the inter- polated record.

The WOCE arrays ACM3 and ACM12 were combined to cover a section almost normal to the slope across the Santos Plateau to the Vema Channel. It was set during Meteor cruise

15/1 in December 1991 and recovered after 23 months later

during Meteor cruise 22/3-4 [Tarbell et al., 1994], providing 685

MOLLER ET AL.: WESTERN BOUNDARY CURRENTS OFF BRAZIL 5431

Table 1. Information on Moorings Off the Brazilian Coast Used in This Analysis

, Water Record

Latitude, Longitude, Depth,

Mooring øN øE m Start Date Days

instrument Depths,

m

VT

BSl/330 -19.487 -038.203 1076 Nov. 14, 1989 368 BS2/331 -19.890 -037.653 1288 Nov. 15, 1989 364

Cabo Frio

CF1/347 -23.081 -040.431 2260 Sept. 20, 1992 325 CF2/348 -22.979 -040.704 980 Sept. 19, 1992 327 CF3/349 -22.948 -040.811 210 Sept. 19, 1992 327 ACM3

BW/333 -27.910 -046.707 1179 Jan. 1, 1991 683 BM/334 -27.987 -046.342 2187 Jan. 1, 1991 682 BE/335 -28.270 -045.230 3258 Jan. 3, 1991 682

775, 1015

100, 300, 700, 1200 173, 375, 880, 1980 240, 920

100

50, 120, 170, 220, 460, 670, 875 280, 530, 930, 1430, 2!37

50, 140, 220, 280, 550, 950, 1450, 2545, 3208

boundary current structure above the level of the AABW [see Tarbell et al., 1994], we restrict this analysis to the three western moorings BW/333 at 1179 m water depth, BM/334 at 2187 m, and BE/335 at 3258 m. Both BW/333 and BE/335 carried

upward looking acoustic Doppler current profilers (ADCP,

array.

For this analysis, we use low-pass filtered (36 hour cutoff period) daily values. For all moorings but BS1/330, pressure records are available from the upper instruments. They are used to determine time series of depths for all current meters in a mooring. These depths were used when calculating trans- ports within water mass layers. The depth ranges of the water masses are taken from conductivity-temperature-depth (CTD) stations obtained during the mooring cruises. They correspond to Zangenberg's [1995] water mass classification in potential temperature salinity space (see Table 2) and encompass the corresponding core densities of extrema that Reid [1989] and rsuchiya et al. [1994] used for their analysis of water mass

circulation.

At array ACM3 the pressure records show some severe drawdowns of the current meters during strong current events [see Tarbell et al., 1994]. Worst case estimates show that the associated errors in speed are negligible (<2 cm/s, or 5%, compared to 40 cm/s for the daily values during an event; <1%

for the record average). Also, the temperature records show that the uppermost instrument never left the layer of the SACW which would be critical for the transport estimates

described below.

The basic statistics of the low-pass filtered daily values are displayed in Table 3. Here SPD and DIR are mean speed and

Table 2. Water Masses as Defined by Zangenberg [1995]

Between Levels of Potential Density

Water Mass Classification

Upper Limit, Lower Limit,

Water Mass kg/m 3 kg/m 3 Depth Interval, m

SACW O-o = 27.10 0-744

AAIW o.o = 27.10 O-o = 27.35 744--1062

UCDW o.o = 27.35 o'2 = 36.70 1062-1317 NADW o'2 = 36.70 o.,• = 45.87 1317-3406

The corresponding depth levels were used for transport calculations.

direction of the current vector with east and north components u and v, respectively, and T is in situ temperature (not mea- sured by ADCPs). The directional stability parameter STAB is

defined as the ratio of the current's vector and scalar means.

STAB = 1 for a current that does not change direction and thus is an indicator for boundary currents that are expected to be directionally stable. Angle brackets denote record averages, and primed quantities are deviations from record means. The timescale is defined as an integral of the autocorrelation func- tion to its first zero crossing. Timescales are calculated for u, v, and T. For the present records, they are of the order of 2{.) days or less, corresponding to 18 degrees of freedom in a year-long record. From the momentum flux (u'v') and the variances (u '2) and (v'2), the main axis (u' v')a of momentum flux is estimated. Although the covariance terms with few ex- ceptions are low, they are presented for completeness.

Transport time series are calculated for the layers of SACW, AAIW, UCDW, and NADW at the WOCE ACM3 array, for the layers of AAIW and NADW at the CF array, and fc)r the flow of AAIW through the western channel of the Vitoria Trindade Ridge. Layer thicknesses were kept constant (see Table 2). For the ACM3 and CF arrays, the widths were de- fined from the distances of the western mooring to the shelf edge, by the halfway distance between neighboring moorings, and by a symmetric extension to the east from the eastern mooring. At the Vitoria Trindade Ridge, the width of the western channel (10 kin) is taken to calculate the transport of AAIW through this channel. Velocities used are the current components normal to the mooring sections (parallel to the channel). Velocities are from meters within the layer and from linearly interpolated values at the layer boundaries. For calcu- lations within the SACW, the velocity at the uppermost current meter is assumed constant to the surface. For mean transports the 95% confidence limits are estimated using the maximum of the integral timescales of the current components involved divided by the record length.

3. Description of Flows and Transports

3.1. Subtropical Layer

The subtropical layer encompasses the depth range from the surface to the lower boundary of the SACW (see Table 2). in Figure 2, all vector time series from this depth range as mea- sured in the three arrays are displayed on the same scale. In the north at the Vitoria Trindade Ridge (20øS), measurements are available only from the eastern mooring. Although this posi-

MOLLER ET AL.: WESTERN BOUNDARY CURRENTS OFF BRAZIL 5433

Brazil Current

100m - 670m

50 0

•-5C

5O 0 -50

BS2

100m

'300•n VT 2O S

0 300 600

Start 15 Nov 89

CF3

0 30o 60o

start 18 Sep 92

0 km BW

220 m 5O

•-5(:

,,

460m 50

-50

,,

' 670rn

50

-50

0 300 600

Start 08 Jan 91

CF2

, .

50

-50

2•,'Om

0 300 600

Start 18 Sep 92

8 km

BM

5O 0 -50

5O o -50

, .

530m

18 km CF1 I

I

t

'" i73rn

•75rn

300 600

Start 19 Sep 92 44 km BE

I ... 50ml

/" 28ornl

50

o T •'.- -•-L .• _-,_

-50

55dm 50

-50

0 300 600 0 300 600

Start 08 Jan 91 Start 08 Jan 91 CF 23 S

ACM3 28 S

60 km 94 km 200 km

Figure 2. Vector time series (northward direction points upward) of low-pass filtered daily mean currents within the depth range of the South Atlantic Central Water (SACW) and the Brazil Current regime, 100-670 m. The plots are arranged from VT (north, top), to CF (middle) to ACM3 (south, bottom), and from east (right) to west (left). Within moorings, lowest levels are top. Distances of mooring sites from the shelf edge (200 m depth contour) are indicated.

ti0n is not in the core of the Brazil Current that may be expected further west on the continental slope [Miranda and

Castro Filho, 1982; Evans et al., 1983], part of the current obviously is deflected by the ridge to the east. This is suggested

0.73) over i year. Also, the current is shallow and does not

reach the 300 m level.

Further south, in the year-long records at Cabo Frio (23øS), the current still is shallow, with mean speeds of about 14 cm/s at both, 100 m depth on the shelf (mooring CF3, STAB = 0.68) and at 173 m depth over the continental break (mooring CF1, STAB = 0.87). The current is replaced at the central mooring

CF2 (240 m depth) by a high stability (0.80) onshore (311 ø) flow. This flow compensates for upwelling off Cabo Frio [Ikeda et al., 1974]. Upwelling events may also be documented by the large temperature flux on the shelf position CFt.

At 28øS (WOCE array ACM3) the Brazil current has in- creased in strength as well as depth. It is confined to the shelf edge (mooring BW), where the current reaches from the sur- face (mean speed up to 44.7 cm/s) down to at least 670 m (6.1 cm/s). Its direction is rather stable (0.64-0.90) for 23 months,

with the few reversals of flow restricted to near the surface. At

larger depths, only one reversal is observed at the beginning of the record; the current keeps its direction for 20 months in its core. The current decreases in strength very quickly at 94 km

20

Tra• [ ACM3:

-40

-60

o 200 400 600 800 o

ACM3: Transport of AAIW

,.t ... ._.,,:.,4.-b t

i ,

200 400 600 800

20 ACM3: Transport of UCDW

> 10-

,m -•o

-20 ' ' '

0 200 400 600

Yearday 1991 800

20 ,,, A,CM, 3: Trans,port o,fN,ADW d

0 200 400 600 800

Yearday 1991

Figure 3. Time series of transports at WOCE array ACM3 within (a) the South Atlantic Central Water (SACW), (b) the Antarctic intermediate Water (AAIW), (c) the Upper Cir- cumpolar Deep Water (UCPDW), and (d) the North Atlantic Deep Water (NADW). The means and standard deviations are indicated by the broken and the dotted lines, respectively;

1 Sv = 10 6 mS/s.

off the shelf edge (mooring BM, 10.3 cm/s at 280 m, 7.9 cm/s at 530 m), and it vanishes almost completely at 200 km off the shelf at mooring BE, where mesoscale eddies dominate and directional stability is low.

Many estimates of the Brazil Current's transport suggest an increase toward the south. Some values reported from geostro- phic calculations are 4 Sv at 20øS [Stramma et al., 1990], around 10 Sv at 23øS and 24øS [Stramma, 1989], 7.3 Sv at 25øS [Campos et al., 1995], 20 Sv at 30øS across the western part of the WOCE section A10 until 43øW [Zangenberg, 1995], and 24 Sv at 34øS [Zemba, 1991]. Part of the increase may to be due to a recir- culation cell [Tsuchiya, 1985; Reid, 1989; Stramma, 1989].

The WOCE moored array ACM3 for 23 months shows the eastern boundary of the Brazil Current between moorings BM

and BE, and maximal flow at the western mooring BW (Figure 2, Table 3). It is unknown how far the Brazil Current extends

In Figure 3a, we show the horizontally integrated transport

ing almost the whole record with exceptions of a few days only.

The 23 month average is -16.2 Sv (poleward), to which the measurements at the western position BW contribute 80%.

The standard error of the average induced by the f!uctuative part of the flow is 2.4 Sv at the 95% confidence limit. In addition, a systematic error must be considered that stems from the assumed geometry, especially from the assumption on the westward extension at BW, where the mean flow is largest.

Reducing the westward extension at BW from the shelf edge to a symmetric one would reduce the poleward transport at BW from -12.6 Sv (see Table 4) to -5.7 Sv, and decrease the total

transport to -9.3 Sv. This value is close to the value (-9.5 Sv)

Fluctuations of transport (1 standard deviation; see Table 4) range from -7 to -26 Sv (Figure 3a). No obvious annual signal can be detected; however, note a timescale much larger than 1 year in the series with the maximum southward trans- port occurring around day 300, even northward transport at the beginning of the series, and low values also at the end of the record (Figure 3a).

3.2. Intermediate Layers

The next two layers encounter the AAIW and the UCDW (Table 2). Starting in the south (28øS), the observations of ACM3 show alternating currents between -20 and 20 cngs near the bottom at BW (Figure 4). The mean is weak (2.0 cm/s)

Table 4. Transport Calculations Where Appropriate in Layers of Water Masses for the Three Arrays VT, CF, ACM3

Transport, Sv Rotation Width,

Array Mooring Angle km SACW AAIW UCDW NADW

VT BS1/330 35 10 0.7

SD 0.3

SE 0.1

CF CF1/347 25 32 0.8

CF2/348 25 22 0.5

Total 62 1.3

SD 0.4

SE 0.1

ACM3 BW/333 20 77 - 12.6 -0.2

BM/334 20 70 -4.2 -1.0 -1.1

BE/335 20 106 0.6 -0.2 -0.9

Total 253 - 16.2 - 1.4 -2.0

SD 9.4 3.4 1.9

SE 2.4 0.5 0.3

-0.5

-0.5 1.6 0.3 -0.3 -2.5 -2.8 4.9 0.8

All transports in Sv (1 Sv - 10 6 m3/s). Standard deviations (SD) and standard errors (SE) due to fluctuations (on the 95% confidence level) are also given. The width of the western channel of the Vitoria Trindade Ridge within the AAIW layer is less (10 km) than that between the 200 m depth contours (14 km).

MOLLER ET AL.: WESTERN BOUNDARY CURRENTS OFF BRAZIL 5435

AAIW, UCDW

700m - 1450m

15 o -15

15

• -15 BS1

775m

1015m

BS2

105• ... 700rn

...-'15 1

0 300 6•0

Start 15 Nov 89

300 600

Start 14 Nov 89

14 km 18 km

BW

I. I 875m

oLJ L&l

0 300 600

Start 08 Jan 91

CF2

0 300 600

start 18 Sep 92 8 km BM

15 930m t

o 'r• I •" •"•t

-15

151' 1430m

0 300 600

Sta• 08 Jan 91

CF1

0 300 600

Start 18 Sep 92 44 km

BE

I 950rn

0 3oo 600

Start 08 Jan 91

60 km 94 km 200 km

Figure 4. Same as Figure 2 but for AAIW and UCDW.

VT 2O S

CF 23 S

ACM3 28 S

and poleward. Further off the shelf edge at BM, the mean flow is poleward in both layers at higher directional stability (about 0.57) and higher mean speed (about 3.5 cm/s). At the deepest mooring (BE), the flow in both layers is still poleward at 1.5 cm/s for the AAIW and 3.5 cm/s for the UCDW. Amplitudes of the speed again reach 15 cmos. Directional stability for the UCDW is higher (0.60) than for the AAIW (0.31). The core of the AAIW poleward flow seems to be close to mooring BM.

Further north, at mooring arrays CF and VT the flow of intermediate waters is equatorward. We observed permanent and directional mostly stable (larger 0.95) equatorward flow of

AAIW with almost no reversals off Cabo Frio. The mean is

slightly higher in the center mooring CF2 than on site CF1

(12.2 crn/s for the interpolated series and 8.6 cm/s, respective- ly). At the Vitoria Trindade Ridge the western channel is deep

10ng record at BS1 shows stable (0.93) and northward flow of AAIW through this channel, with extremely high mean speed

of 20.8 cm/s. In the second channel at BS2 the situation is not

that pronounced. The mean flow of AAIW seems to pass the channel equatorward (309 ø due to topography) at 6.1 cm/s;

however, the variability is high and stability is relatively low.

At ACM3 the predominant flow of both the AAIW and the UCDW is poleward. The resulting transport time series (Fig- ures 3b and 3c) show higher variability in the AA!W layer than

in the UCDW layer. The central mooring contributes most of the transport within the AAIW. The mean transport is south- ward in both layers, with a contribution of -1.4 Sv from the AAIW and -2.0 Sv from the UCDW (Table 4). The total of -3.4 SV (AAIW and UCDW) with low standard error (0.5 Sv) compares well with a result from a geostrophic calculation [Zangenberg, 1995] of -4 Sv at 30% west of 43øW from WOCE

section A10.

At Cabo Frio the transport of AAIW is permanently north- ward (Figure 5a; Table 4), with 1.3 Sv on average (0.1 Sv standard error) and with a standard deviation of 0.4 Sv. In the western channel of the Vitoria Trindade Ridge, an average northward transport of 0.7 Sv was observed (standard error 0.1 Sv; Figure 5c). As the average transports at Cabo Frio and through the channels of the VT regions should be similar, it is probable that part of AAIW also passes through the eastern channels of the Vitoria Trindade Ridge. However, the mea-

surements at BS2 in the second channel are too variable and

too short to give a reasonable transport estimate for that chan-

nel.

3.3. North Atlantic Deep Water

Flow of NADW was observed off Cabo Frio and in the

ACM3 array (Figure 6). Off Cabo Frio at CF1, we find NADW flowing southward in the mean (4.3 cm/s) with high variability

3 CF: Transport of AAIW

400 600

BS, I: Transport of A•,IW

400 6•0

Yearday 1989 CF: Transport of NADW

5 ' b

0

-5 I' i

400 600

Yearday 1992

Figure 5. Same as Figure 3 but for the CF array within (a) the AAiW and (b) the NADW and (c) in the western channel of the Vitoria Trindade Ridge within the AAIW.

(-+15 cm/s) and relative low stability (STAB = 0.53). From

layer that must be deflected to the east by the ridge system.

At ACM3 the NADW flows poleward in the mean. How- ever, as at Cabo Frio, the flow is variable and the resulting poleward transport is weak (-2.8 Sv, with 4.9 Sv standard

deviation and 0.8 Sv standard error, Figure 3d, Table 4).

Geostrophic transport estimates at 30øS from WOCE section

A10 give -10 Sv west of 41øW [Zangenberg, 1995], which due

whole western basin [Zangenberg and Siedler, this issue]. None

the Santos Plateau until the Vema Channel, show a pro- nounced permanent poleward flow within the NADW layer

CF1

NADW • •oS• ,08 OF

1980m- 2545m o -15[ ' '•"'

23 S

0 300 600

Start 18 Sep 92 44 km

BM BE

15 r ... 2137ml I' ' 2545ml

-15 -15

0 300 600 0 300 600

Start 08 Jan 91 Start 08 Jan 91

94 km 200 km

Figure 6. Same as Figure 2 but for the NADW.

of a we!l-developed western boundary current. It may even not reach the Malivinas and Brazil Current Confluence Zone [Weatherly, 1993]. Zangenberg and $iedler [this issue] argue that conservation of potential vorticity in the presence of the Vito- ria Trindade Ridge and the Santos Plateau forces the NADW to leave the shelf break and to flow eastward and northeast.

ward.

4. Conclusions

The direct measurements in three moored current meter arrays between 20øS and 28øS on the Brazilian continental slope establish the deepening and strengthening of the Brazil Current from 20øS to 28øS. At 28øS the records of 23 months' length show that it is a permanent boundary current that reaches down to more than 670 m, with an estimate of 16 Sv poleward transport west of 45øW (position BE).

For the layers of the AAIW and UCDW the direct measure- ments correspond well with the circulation pattern for the AAIW that was suggested from geostrophic and water mass

analysis by Reid [1989] and confirmed later by $uga and Talley

[1997] achieved from an analysis of WOCE float trajectories.

The question that was posed by Warner and Weiss [1992] in

their final figure for the path of the AAIW circulation off CF can be answered from direct measurements. The AAIW and the UCDW enter the southern Brazil Basin from the east

[Boebel et al., 1997] and split into two branches along the continental slope. One branch turns poleward and passes

directed equatorward already at 25øS [Campos et al., 1996], and

through the western channel of the Vitoria Trindade Ridge.

Recently repeated measurements off Cabo Frio confirm the equatorward flow of intermediate waters (J. Lima and J. H.

Middleton, personal communication, 1997).

Although a deep western boundary current carrying NADW is well established north of the Vitoria Trindade Ridge [Harkema and Weatherly, 1996], we were not able to identify it as clearly south of the ridge. It may exist on the slope at Cabo Frio east of array CF. At 28øS (ACM3), it is only weakly developed, with much less mean poleward transport (-2.8 Sv) than expected from geostrophic estimates [Zangenberg and $ie- dler, this issue], and with several reversals toward north. It remains unclear from the direct measurements along 28øS in ACM3 and ACM12 [see Tarbell et al., 1994] if and where the NADW is transported southward in a more pronounced west- ern boundary current.

Acknowledgments. We appreciate the technical assistance of the mooring groups of IFMK and IOUSP. We would also like to thank the captains and crews of the FS Meteor and N/Oc. Prof. W. Besnard for their skilled work at sea. Two reviewers gave valuable comments. This study was supported by the Deutsche Forschungsgemeinschaft (FS Meteor cruises Mll/2, M14/3, M15/1, M22/3) and the Bundesminister ffir Forschung und Technologie (Az 03FO157A), both in Bonn, Ger- many; by FUNDESPA/PETRoBRAS (N/Oc. Profi W. Besnard cruises in 1992 and 1993), Sao Paulo, Brazil; by CNPq, Brasilia; and by the Brazilian-German bilateral program in marine science and technology (MAR 14 and MAR 25).

References

Boebel, O., C. Schmid, and W. Zenk, Flow and recirculation of Ant- arctic Intermediate Water across the Rio Grande Rise, J. Geophys.

Res., 102(C9), 20,967-20,986, 1997.

MOLLER ET AL.: WESTERN BOUNDARY CURRENTS OFF BRAZIL 5437

Buscaglia, J. L., On the circulation of Intermediate Water in the southwestern Atlantic, J. Mar. Res., 29(3), 245-255, 1971.

Campos, E. J. D., J. E. Goncalves, and Y. ikeda, Water mass charac-

Campos, E. J. D., Y. Ikeda, B. M. Castro Filho, S. A. Gaeta, J. A.

Lorenzzetti, and M. R. Stevenson, Experiment studies circulation in the western South Atlantic, Eos Trans. AGU, 77(27), 253, 259, 1996.

Defant, A., Schichtung und Zirkulation des Atlantischen Ozeans, Die Troposphfire, Wiss. Ergebn. Dtsch. Atl. Exped. Meteor 1925-1927, 6(3), part 1, 289-411, 1936a.

Defant, A., Ausbreitungs- und Vermischungsvorg•inge im Antarktis- chen Bodenstrom und im Subantarktischen Zwischenwasser, Wiss.

Ergebn. Dtsch. Atl. Exped. Meteor 1925-1927, 6(2), part 2, 75-96,

1936b.

I)eMadron, X. D., and G. Weatherly, Circulation, transport and bot- tom boundary layers of the deep currents in the Brazil Basin, J. Mar.

Res., 52, 583-638, 1994.

Evans, D. L., and S. R. Signorini, Vertical structure of the Brazil Current, Nature, 315, 48-50, 1985.

Evans, D. L., S. R. Signorini, and L. B. Miranda, A note on the transport of the Brazil Current, J. Phys. Oceanogr., 13, 1732-1738,

1983.

Harkema, R., and G. L. Weatherly, A compilation of moored current meter data from the western boundary of the Brazil Basin for the Deep Basin Experiment, September 1993-March 1995, Tech. Rep.

CMF-96-01, 72 pp., Dep. of Oceanogr., Fla. State Univ., Tallahassee,

1996.

Hinz, I<2, L. Hasse, and F. Schott, Subtropischer und tropischer At- lantik, Reise 14/!-3, Maritime Meteorologie und Physikalische Oz- eanographie, 17 September-30 December 1990, Meteor Ber. 91-3, 58 pp., Univ. Hamburg, Hamburg, Germany, 1991.

Hogg, N. G., and W. Zenk, Long-period changes in the bottom water flowing through Vema Channel, J. Geophys. Res., !02(C7), 15,639-

!5,646, 1997.

Hogg, N. G., P. E. Biscaye, W. Gardner, and W. J. Schmitz Jr., On the transport and modification of Antarctic Bottom Water in the Verna Channel, J. Mar. Res., 40, suppl., 231-263, !982.

Holfort, J., GroBrfiumige Zirkulation und meridionale Transporte im Siidatlantik (with English abstract and figure captions), Bet. Inst.

Meereskd. Univ. Kiel, Kiel, 260, 96 pp., 1994.

Ikeda, Y., L. B. Miranda, and N.J. Rock, Observations of stages of upwelling in the region of Cabo Frio (Brazil) as conducted by con- tinuous surface temperature and salinity measurements, Bol. Inst.

Oceangr. Univ. Sao Paulo, 23, 33-46, 1974.

Miranda, L. B., and B. M. Castro Filho, Geostrophic flow conditions of the Brazil current at 19øS, Cienc. Interam., 22, 44-48, 1982.

Miranda, L. B., A. S. Mascarenhas Jr., Y. Ikeda, T. A. Rago, and P. L.

Cacciari, Resultados preliminares da estrutura termica e do campo de velocidade amostrados durante o cruzeiro oceanografico Transcobra IIi, Relat. Cruzeiros Ser. N/Oc. Prof. W. Besnard, 6, 1-13,

1985.

011itraut, M., N. Cortes, G. Laec, and J.-P. Rannou, Maryor float present results from the SAMBA.experiment, Oceans '94 Proc., III, 17-22, 1994.

Peterson, R. G., and L. Stramma, Upper level circulation in the South Atlantic Ocean, Prog. Oceanogr., 26, 1-71, 1991.

Reid, J. L., On the total geostrophic circulation of the South Atlantic Ocean: Flow patterns, tracers, and transports, Prog. Oceanogr., 23, 149-244, 1989.

Reid, J. L., W. D. Nowlin Jr., and W. C. Patzert, On the characteristics and circulation of the southwestern Atlantic Ocean, J. Phys. Ocean- ogr., 7(1), 62-91, 1977.

Schott, F., J. Fischer, J. Reppin, and U. Sen•, On mean and seasonal

currents and transports at the western boundary of the equatorial

Atlantic, J. Geophys. Res., 98(C8), 14,353-14,368, 1993.

Spain, P. F., D. B. Dorson, and H. T. Rossby, PEGASUS: A simple, acoustically tracked, velocity profiler, Deep Sea Res., 28A(12), 1553- 1567, 1981.

Speer, K. G., and W. Zenk, The flow of Antarctic bottom water into the Brazil Basin, J. Phys. Oceanogr., 23, 2667-2682, 1993.

Stommel, H., The westward intensification of wind-driven ocean cur- rents, Eos Trans. AGU, 29(2), 202-206, 1948.

Stommel, H., and A. B. Aarons, On the abyssal circulation of the world ocean, ii, An idealized model of the circulation pattern and ampli- tude in oceanic basins, Deep Sea Res., 6, 2!7-237, 1960.

Stramma, L., The Brazil Current transport south of 23øS, Deep Sea Res., 36, 639-646, 1989.

Stramma, L., Y. Ikeda, and R. G. Peterson, Geostrophic transport in the Brazil Current region north of 20øS, Deep Sea Res., 37, 1875- 1886, 1990.

Suga, T., and L. D. Talley, Antarctic Intermediate Water circulation in the tropical and subtropical South Atlantic, J. Geophys. Res., 100(C7), 13,441-13,453, 1995.

Taft, B. A., Distribution of salinity and dissolved oxygen on surfaces of uniform potential specific volume in the South Atlantic, South Pa- cific, and Indian Oceans, J. Mar. Res., 21(2), 129-146, 1963.

Tarbell, S., R. Meyer, N. Hogg, and W. Zenk, A moored array along the southern boundary of the Brazil Basin for the Deep Basin Experiment--Report of a joint experiment 1991-1992, Tech. Rep.

94-07, Woods Hole Oceanogr. Inst., Woods Hole, Mass., 1994. (Also Ber. Inst. Meereskd. Univ. Kiel, 243, 97 pp., 1994.)

Tsuchiya, M., Evidence for a double-cell subtropical gyre in the South Atlantic Ocean, J. Mar. Res., 43(1), 57-65, 1985.

Tsuchiya, M., L. D. Talley, and M. S. McCartney, Water-mass distri- bution in the western South Atlantic, J. Mar. Res., 52, 55-81, 1994.

Warner, M. J., and R. F. Weiss, Chlorofluoromethanes in the South Atlantic Antarctic Intermediate Water, Deep Sea Res., 39, 2053- 2075, 1992.

Weatherly, G. L., On deep-current and hydrographic observations from a mudwave region and elsewhere in the Argentine Basin, Deep Sea Res. H, 40(4/5), 939-961, 1993.

Whitworth, T., III, W. D. Nowlin Jr., R. D. Pillsbury, M. I. Moore, and R. F. Weiss, Observations of the Antarctic Circumpolar Current and Deep Boundary Current in the southwest Atlantic, J. Geophys. Res., 96(C8), 15,105-15,118, 1991.

Wrist, G., Schichtung und Zirkulation des Atlantischen Ozeans. Das Bodenwasser und die Stratosph•ire, Wiss. Ergebn. Dtsch. Atl. Exped.

Meteor 1925-I927, 6(2), part 1, 1-288, 1935.

Zangenberg, N., Die Zirkulation des Oberfl•ichen und des Tiefenwas- sers im Sfidlichen Brasilianischen Becken (English abstract and fig- ure captions), Ber. Inst. Meereskd. Univ. Kiel, 28.!, 142 pp., 1995.

Zangenberg, N., and G. Siedler, The path of the North Atlantic Deep Water in the Brazil Basin, J. Geophys. Res., this issue.

Zemba, J. C., The structure and transport of the Brazil Current be- tween 27øS and 36øS, Ph.D. thesis, 164 pp., Woods Hole Oceanogr.

Inst. Tech. Rep. 91-37, Woods Hole Oceanogr. inst., Woods Hole, Mass./Mass. Inst. of Technol., Cambridge, 1991.

Zenk, W., and N. G. Hogg, Warming trend in the Antarctic Bottom Water flowing into the Brazil Basin, Deep Sea Res., 43, 1461-1473,

1996.

Y. !keda and L. V. Nonato, Instituto Oceanografico da Univer- sidade de Sao Paulo, Praca do Oceanografico, 191, Cidade Universi- taria, 05508-900 Sao Paulo, SP, Brazil.

T. J. Mfiller and N. Zangenberg, Institut ffir Meereskunde an der Universitat Kiel, Dfisternbrooker Weg 20, 24105 Kiel, Germany. (e- mail: tmueller@ifm.uni-kie!.de)

(Received September 27, 1996; revised September 11, 1997;

accepted October 16, 1997.)